There is a server system in which when there is a high load level for service requests from a client, the service request is fulfilled by adjusting the number of processes for applications. The server system adjusts the number of processes for applications on the basis of a response time for a service request and the number of service requests estimated in advance per unit of time.
The number of processes for applications may be manually adjusted by a system administrator. A server system capable of monitoring the state of the load due to service requests and automatically adjusting the number of processes has been disclosed. Japanese Laid-open Patent Publication No. 2006-317991 and Japanese Laid-open Patent Publication No. 9-265409 disclose examples of the related art.
The number of processes is desirably adjusted in accordance with the peak load of service requests. It is, however, difficult to appropriately estimate the peak load. The cost of starting and stopping processes is high, and it is difficult to follow sudden changes in load. Therefore, when adjusting the number of processes, resources are still needlessly consumed for much of the time thereof.